<![CDATA[According to data from the 2019 National Socio-Economic Survey (Susenas), there are 26 million people in Indonesia with disabilities or around 9.7% of the total Indonesian population. People with physical disabilities who have difficulty walking require a wheelchair to move around. Some causes of physical disabilities include multiple disabilities in the legs and arms, accidents, quadriplegia, and stroke. An automated wheelchair is a solution to this problem, which is controlled without physical touch and uses the YOLO (You Only Look Once) object detection algorithm to detect the direction of motion of the user's head. The results of several epochs conducted show that YoloV5 small has an f1 score of 0.9 and a smaller loss, so it will be used in the next test. The system created is able to move according to the input given, with a computation time using CUDA of around 65 milliseconds which is relatively fast. Lower power usage and faster computation time are also evident when using CUDA, although there is a discrepancy in the CPU utilization values of cores 2 and 3 which stay at 0%.]]>
Copyrights © 2022
